What can wastewater tell us about an outbreak?

Most outbreak data depends on a chain of decisions: someone feels unwell, decides to seek care, is offered a test, and the result is reported. Every link loses information, and each is affected by access, cost and awareness. Wastewater surveillance skips the chain.

Measuring everyone at once

People infected with many pathogens shed detectable material in stool or urine — often starting before symptoms and continuing in people whose infection is mild or asymptomatic. That material arrives at a treatment plant, mixed. Sampling there gives a measurement across the whole population connected to the sewer, without anyone having to present themselves anywhere.

A study published in Nature Biotechnology in 2020 demonstrated that SARS-CoV-2 RNA concentrations measured in primary sewage sludge tracked community infection dynamics, and did so ahead of some clinical indicators.

What it is good at

Direction and timing. Whether transmission is rising or falling, and roughly when the change began. This is the core strength, and it holds even when testing behaviour changes.

Coverage independent of care-seeking. A signal that does not depend on who can afford, access or be bothered to get tested. When clinical testing declines — as it does once an outbreak stops being novel — wastewater does not go blind in the same way.

Detecting something before it is looked for. Because samples can be tested for a target that was not being monitored clinically, wastewater can reveal presence in a community without a single reported case. This is how it has been used for poliovirus, where most infections cause no symptoms at all and waiting for a paralysed child means waiting until transmission is well established.

Reading the pathogen, not just counting it. Sequencing the material recovered can show which variants are present in a community, again without depending on who happened to have a sample taken.

What it is poor at

Counting cases. Turning a concentration into a number of infected people requires assumptions about how much each person sheds, for how long, dilution, rainfall, temperature, and how long material takes to arrive in the sewer. Each is uncertain and they compound. Treat case estimates from wastewater with caution.

Saying anything about an individual. The sample is a mixture. It contains no information about any specific person, and this is a property of the method rather than a safeguard added to it.

That protection weakens as the catchment shrinks. Sampling a whole city reveals nothing about anybody; sampling a single building can, in principle, point at a small and identifiable group. The ethical question is not about the technique but about where the sampling point is placed, and it deserves to be settled before a programme starts rather than after.

Covering everyone. It measures people connected to the sewer being sampled. Households on septic systems are absent, as are populations served by other infrastructure — and those are often the communities already least visible in health data. A method praised for not depending on care-seeking still depends on plumbing.

Making two measurements comparable

A raw concentration is affected by how much water is in the system. Heavy rain dilutes everything; industrial discharge changes composition; flow varies through the day and week. A signal that halves after a storm has not necessarily told you anything about infection.

Programmes therefore normalise — adjusting for measured flow, or for the concentration of some other marker expected to be roughly constant per person. One widely used marker is a plant virus abundant in human stool because of what people eat, which serves as a rough proxy for how much human waste is in a sample.

Whether this actually improves the correlation with clinical data is itself an active question: a 2023 study in Science of the Total Environment examined exactly that and found the benefit less clear-cut than assumed. This is worth knowing, because normalisation is often presented as a solved detail when it is a live methodological argument.

What happens between the sewer and the number

The path from sample to data point has several steps that each introduce variation: concentrating a small amount of genetic material out of a large volume of dirty water, extracting it, and quantifying it. Recovery is never complete, and how incomplete depends on the method.

This is why laboratories run controls to estimate how much they lost, and why two laboratories analysing the same sample can report different absolute numbers while agreeing on the trend. It is another argument for reading direction rather than magnitude, and for being suspicious of any comparison that crosses a laboratory boundary.

Reading a wastewater chart honestly

The useful question is "which way is this moving, and since when", not "how many people does this represent". Compared against a plant's own history the signal is informative. Compared between plants with different catchments, industrial inputs, rainfall and laboratory methods it is much less so, and cross-city league tables built from wastewater data should be read with real scepticism.

Used that way — as an early, population-level trend indicator rather than a count — it adds something no clinical testing system provides: a measurement that does not care whether anybody decided to seek care.